Lean Research

Goal & hypotheses

Question: Which database fits a single-region internal inventory app with about 20 concurrent users?

• H1: SQLite is sufficient if writes are brief and serialized, with one application host accessing the database file.
• H2: Postgres is the safer default if concurrent writes, multiple application hosts, or database-level access control are expected.

Findings

• SQLite permits multiple readers, but only one writer at a time. This fits low-write workloads when transactions are short. Source: SQLite, “Appropriate Uses For SQLite,” https://www.sqlite.org/whentouse.html
• SQLite’s own guidance recommends a client/server database when many programs access the database over a network or when many concurrent writers are needed. Source: SQLite, “When To Use SQLite,” https://www.sqlite.org/whentouse.html
• PostgreSQL uses a client/server architecture and supports concurrent sessions with transaction isolation and row-level locking. Source: PostgreSQL documentation, “Concurrency Control,” https://www.postgresql.org/docs/current/mvcc.html

Contradictions & gaps

“20 concurrent users” does not establish write concurrency. Twenty people mostly reading and occasionally saving short updates can have a different load from 20 simultaneous bulk imports. Hosting topology, backup requirements, and operator experience are also unspecified. Documentation describes capabilities, not this app’s measured workload.

Conclusion

Start with SQLite if the app runs on one host, writes are short and infrequent, and the team can tolerate brief write contention. Choose Postgres if you expect multiple app hosts, frequent concurrent writes, or need a separately managed database service. Before committing, test the busiest expected workflow and confirm backups and restore procedures.

Next steps

1. Measure simultaneous writes in a representative import/update test.
2. Confirm whether deployment will use one application host.
3. Verify backup restoration for the selected database.